Figure 1 Spectrum of biological rhythms in a human body. The spectrum of circadian rhythms is presented in this figure at the level of period duration of 1 day (log 105), indicated by rhythms of ‘sleep-wakefulness’ (left side) and ‘earth rotation (right side) (Figure taken from Moser et al., 2006 )...3 Figure 2 The structure of sleep with the 5 different stages, REM-stage and NREM-stages I to
IV. These 5 stages constitute one sleep cycle that takes between 90 and 110 minutes. The duration of the REM-sleep stages increases over the course of the sleep period (adapted from Roenneberg, 2006 ). ...5 Figure 3 Two-process model of sleep regulation by Serge Daan und Alex Borbély (Borbély
(1982) and Daan (1984) ). The interplay between these two processes regulates both
135
timing and duration of the sleep period (I, without sleep deprivation). After sleep depression the “slow wave power” at the beginning of sleep period increases and the duration of the recreation sleep is prolonged (II, with sleep deprivation) (adapted from Roenneberg, 2006 )...7 Figure 4 Image of the anatomical connections between the eye (light reception at the retina)
and the neuronal pathway via the optic nerves to the suprachiasmatic nucleus (SCN). The figure additionally shows that Melatonin (synthesised in the pineal gland) is secreted into the blood system (source http://thebrain.mcgill.ca)...9 Figure 5 Distribution of Chronotypes calculated by mid-sleep on free days (MSF) from our
MCTQ Database with entries from 60.000 people mainly dwelling in Germany,
Switzerland, Austria and The Netherlands...12 Figure 6 Distributions of Chronotypes judged by different calculations of mid-sleep. The
figure on top shows the simple mid-sleep on free days (MSF). The figure in the middle shows the MSF corrected for the sleep debt accumulated during the workweek (MSFsc, see text for details). The figure at the bottom shows the MSFsc further corrected for age- and sex-dependent changes (MSFsasc, see text for details). (Taken from Roenneberg et al., 2007a )...13 Figure 7 Analysis of the original version of the Munich ChronoType Questionnaire (MCTQ)
identified the rudimentary questions, necessary for quantitative Chronotype assessment.
To avoid any confusion concerning the individual questions (e.g., when do you go to bed, get ready to fall asleep, etc.), cartoons exemplify the sequence of events from the time people go to bed and get up. Subjects filled out the MCTQ at the onset of each study period. ...25 Figure 8 Sleep Log: Subjects filled out a sleep-log every morning after wake-up. Its questions
relate to those in the MCTQ (Figure 7). ...26 Figure 9 Image of a wrist worn actimetry device (Daqtometer Version 2.3). An integrated
dual axis accelerometer (not shown) records both dynamic (motion) and static (gravity, i.e. change in position) acceleration. The energy source is a standard 3 Volt watch battery (CR2032)...27 Figure 10 Seasonality in sleep timing taken from the MCTQ database (n 55,000) Annual
time courses are double plotted (the same data is shown sequentially to more easily visualize systematic trends). A. Half-monthly averages of Mid-Sleep times on Free days , MSF (open circles ± SEM) and of wake-up times (line). DST-periods are indicated by the open boxes and their transitions by stippled horizontal lines; dawn times are shown as a grey to white border. Whereas sleep times track dawn under standard time, mid-sleep is scattered around 3:30 (wake-up times around 7:40) under DST. Age and sex ratio were not significantly different in the 24 averages and showed no interactions.
B. Seasonal changes in sleep duration (averaged over both free and work days) result in about 20 min more sleep in winter than in summer (cosine fit: r = 0.75; p<0.0001). ...29 Figure 11 Comparison of sleep times and activity profiles between different Chronotypes.
Sleep times (black bars) and activity (black lines) – recorded during the two longitudinal studies around the autumn and the spring DST-transition – averaged for the free days within the four weeks before the autumn change in an early (top) and a late (bottom) Chronotype. Sleep-onset and -offset times are taken from the sleep-logs; activity levels were measured by wrist actimetry. The phase of mid-sleep is indicated by an open circle within the sleep bar and the phase of the Centre of Activity (CoAct, see Methods) as a black square. Chronotype correlated highly with the CoAct at baseline (see Methods;
r = 0.56, p<0.0001). Sleep log entries also correlated with the sleep-times extracted from the activity records similarly for both transitions (for the autumn: sleep-onsetbefore: r = 0.38, p<0.001; sleep-endbefore: r = 0.7, p<0.001; sleep-onsetafter: r = 0.22, p<0.005; sleep-endafter: r = 0.55, p<0.001)...30
136
Figure 12 Adjustments to DST-transitions of sleep and activity times resulting from the longitudinal study (n=50) A. Phase adjustments of mid-sleep (circles) and activity
(CoAct, black squares) around the DST-transitions expressed as weekly averages relative to each individual’s baseline (average phase during the four pre-transition weeks, see Methods). Results are shown for the entire cohort both on free (left panels) and on work days (right panels). The autumn transition is shown in the top panels, the spring
transition in the bottom panels. Horizontal bars connected to the respective symbols represent SEM which were in most cases smaller than the size of the symbols. B. The comparison between early (left panels) and late Chronotypes (left panels) is shown for free days only (otherwise as in A). For the changes of mid-sleep on free days in autumn, a mixed ANOVA (within-subject design with Chronotype, early, intermediate and late, as a between-subject factor) shows a significant difference between all weeks (F (4.33;
117) =10.00, p<0.001). For both transitions, post-hoc tests show that neither the 4 pre- nor the four post-transition weeks differ among each other, while they differ significantly across the transitions. In autumn, the CoAct times show no difference between the 8 weeks (F (3.5; 94) =1.89, p=.13). The changes for CoAct of early Chronotypes correlates better with dawn than with social time (r: 0.938 vs. 0.896). In spring, the phase changes of both mid-sleep and CoAct differ significantly before versus after the transition (mixed design ANOVA; mid-sleep: F (4.57; 128) =20.26, p.001; CoAct: F (4.84; 170) =4.36, p.001) while they are statistically indifferent among the pre- and post-transition weeks.
The changes for CoAct of late types between week 1 and 6 correlate better with dawn than with social time (r: 0.974 vs. 0.774). Whereas post-hoc tests show that the final phases reached in the last two weeks show no significant difference relative to any of the 4 weeks prior to transition for both Chronotypes, they differ significantly between early and late types (t(49)=2.13, p0.05). ...31 Figure 13 Relationship between natural and behavioural light: dark cycles (with and without
DST). The relationship between the natural light-dark cycle (dawn, zeitgeberN; solid curve) and the behavioural light-dark cycle (created by using artificial light and sleeping in dark rooms with closed eyes, zeitgeberB, exemplified by an arbitrary wake-up time at 7 a.m.; dotted line) changes systematically with season (A). DST only affects zeitgeberB
by advancing the social clock by one hour in spring and delaying it in autumn (B). The 1-hour advance corresponds to travelling 15° westward within the same time zone. DST-transitions have large effects on the seasonal relationship between the two zeitgebers.
This phenomenon becomes more apparent if natural dawn is drawn with respect to local time (consistent with social wake-up times) (C). The seasonal progression of the phase relationship between the two zeitgebers is delayed by 4 weeks in the spring and by 6 weeks in the autumn (vertical grey arrows). Hence, we repeat almost 20% of the seasonal progression of the two zeitgebers every year. In addition, DST artificially changes the amplitude of the phase relationship in summer (horizontal white arrows in B and C), which mimics a translocation of 17° latitude. The diagrams are drawn for the dawn times in Frankfurt/Main (50°7’N/8°41’E) which roughly corresponds to the average coordinates of the 50 subjects’ places of residence. In this case, the longitudinal and latitudinal translocations would mean moving from Frankfurt to Morocco in spring and back in autumn. The amplitude of the relationships as well as the degree of their perturbations by DST increases with latitude...34 Figure 14 Distribution (in percent) of shift-workers within the 27 EU and two EFTA
(European Free Trade Association, CH, NO) countries (taken from the Fourth European Working Conditions Survey, European Foundation for the Improvement of Living and Working Conditions, 2007)...36 Figure 15 Distribution (in percent) of shift-workers by work sector within the 27 EU and two
EFTA (European Free Trade Association, CH, NO) countries (taken from the Fourth
137
European Working Conditions Survey, European Foundation for the Improvement of Living and Working Conditions, 2007). ...36 Figure 16 Flow chart of the internet literature search. On average 1995 hits have been
identified from the search on the terms ‘shift-work / night work / night shift / shift
schedule / alternating shift / alternating night shift / rotating shift / rotating night shift’. A final number of 52 articles remained for further analyses. ...44 Figure 17 Distribution of nations from the selected 52 articles. The majority of studies have
been performed in Japan, the United States of America and Sweden. ...46 Figure 18 Distribution of study types in the selected 52 articles. Most studies are
cross-sectional studies (57%), followed by similar equal numbers of retrospective and
prospective studies (24% and 19%, respectively). ...46 Figure 19 Distribution of sexes/genders studied in the selected 52 articles. The majority of
studies investigated exclusively male subjects (43%, ‘males only’), followed by 30% of studies with female subjects only. One fifth (22%) of the studies did not give the
information on the female/male ratio (‘sexes not separated’). Only 5% of the studies listed the results separately for females and males...49 Figure 20 Distribution of the selected 52 articles due to the 4 main health categories. The
majority of articles publishes results on cardiovascular problems (31%), followed by sleep and metabolism (with each 25%) and finally 19% on cancer diseases...52 Figure 21 Taxonomy of the selected articles about sleep problems (n=13). The taxonomy
shows the distribution of articles showing significant () and non-significant (Ø)
differences in sleep problems between the shift-work and control group for the respective work schedule, shift rotation, sexes and occupational group...57 Figure 22 Taxonomy of the selected articles about cardiovascular problems with Coronary
Heart Disease and Hypertension (n=16). The taxonomy shows the distribution of articles showing significant () and non-significant (Ø) differences in cardiovascular problems between shift-work and control group, for the respective work schedule, rotation, sexes and occupational group. ...63 Figure 23 Taxonomy of the selected articles about breast, colorectal, endometrial and prostate
cancer (n=10). The taxonomy shows the distribution of articles showing significant increases in cancer for the respective work schedule, rotation, sexes and occupational group. ...75 Figure 24 Taxonomy of the selected articles about metabolic problems with ulcer and
duodenitis, diabetes and subfertility (n=13). The taxonomy shows the distribution of articles showing significant () and non-significant (Ø) differences in metabolic problems between the shift-work and control group for the respective work schedule, rotation, sexes and occupational group. ...79 Figure 25 Illustration of the main health effects identified from the selected articles, grouped
by shift schedule and direction of rotation. Most findings have been published for clockwise 3-shift systems with 8 hour shift durations. The least number of results has been found for 5x8 h schedules (5-shift systems with 8 hour shifts) and 4x6 h schedules (4shift systems with 6 hour shift). ...86 Figure 26 Illustration of the interplay of social schedule and biological zeitgebers and
consequences of a mismatch between these. Shift-work (indicated by a red flash) represents a social factor that disturbs this interplay. The two possible results from this disturbance are ‘stress and disease’ (left side) or ‘adapted circadian rhythms’ (right side).
...91 Figure 27 Distribution of the 4 major health categories sleep (blue bars), cardiovascular (red
bars), metabolic (green bars) and cancer problems (purple bars). The results are presented by the shift rotations, namely clockwise (forward), counterclockwise
138
(backward) and fixed night work. The results from studies that did not name the direction of shift rotation are grouped under ‘not defined’. ...96 Figure 28 Distribution of the 2 most consistent health outcomes sleep and cardiovascular
problems. The results are presented by the different shift rotations, namely clockwise (forward), counterclockwise (backward) and fixed night work. The results from studies that did not name the direction of shift rotation are grouped under ´not defined´. ...96 Figure 29 The figure shows the potential health costs (in billion Dollar) arranging from
shift-work on the development of Coronary Heart Disease (CHD). A trend can be seen that costs increase with shift-work seniority up to 10 years and then decrease for longer time in shift-work. For explanations on this finding see text. The table below the figure shows total costs in billion Dollars and percentage costs from grant total 156.4 billion Dollars.
...103 Figure 30 Screenshot of the Shift-Work/Social-Jetlag Model (Roenneberg et al., in
preparation). The main functions are indicated, with the options to enter Chronotype (A) as an input variable, to set the advance and delay capacities (B), and to set the start and end times of the work shifts (C). Additionally the program contains a data table (D) with information about the resultant values of Social Jetlag. The corresponding double-plot (E) graphically shows the relationship between internal time (green circles) and external time (red squares, given by the work hours) for one shift-cycle. ...109 Figure 31 Two double-plots yielded from the Shift-Work/Social-Jetlag-Model (Roenneberg et
al., in preparation) showing the differences between an early (left figure) and late Chronotype (right figure). It is conspicuous, that the early Type (left) on the early shift (red area) does not accumulate Social Jetlag, whereas the late (right) Type does. In turn, on the night shifts (blue areas) the late Type does not accumulate that much Social Jetlag as the early Type...110 Figure 32 Distribution of MSF in the shift-work population of the Automobile-Test-Sample.
The distribution is slightly skewed rightwards; with a higher number of late
Chronotypes. The average Chronotype of the sample is 4.7...112 Figure 33 Calculation from the Shift-Work/Social-Jetlag-Model for the comparison of the
levels of Social Jetlag, accumulated in one shift cycle in a clockwise (blue bars) and a counterclockwise (red bars) rotational shift cycle. Clockwise shift rotation always leads to significantly elevated levels of Social Jetlag, for all three Chronotypes (** p<0.001 /
*** p<0.0001; Early p=0.0009, Intermediate p<0.0001, Late p<0.0001; Mann-Whitney-U-test)...113